Re: [Apbs-users] Comparing APBS PB Grid with Congen Grid?
Biomolecular electrostatics software
Brought to you by:
sobolevnrm
Menu
▾
▴
Re: [Apbs-users] Comparing APBS PB Grid with Congen Grid?
|
From: Chris C. <ccu...@so...> - 2007-04-10 16:21:46
|
Hi Nathan,
Below are two input files that gave us some of the best results that
David referred to in the last e-mail.
read
mol pqr 1rne.pqr
end
elec
mg-auto
dime 161 161 161
cglen 217.6 217.6 217.6
fglen 128 128 128
cgcent mol 1
fgcent mol 1
mol 1
npbe
bcfl mdh
pdie 4.0000
sdie 78.0000
ion 1 0.15 2.0
ion -1 0.15 2.0
srfm smol
chgm spl2
sdens 15.00
srad 1.40
swin 0.3
temp 300.0
calcenergy no
calcforce no
write pot dx 1rne
end
quit
read
mol pqr 1rne.pqr
end
elec
mg-manual
dime 161 161 161
nlev 4
glen 128 128 128
gcent mol 1
mol 1
npbe
bcfl mdh
pdie 4.0000
sdie 78.0000
ion 1 0.15 2.0
ion -1 0.15 2.0
srfm smol
chgm spl2
sdens 15.00
srad 1.40
swin 0.3
temp 300.0
calcenergy no
calcforce no
write pot dx 1rne
end
quit
Thanks,
Chris
On Apr 9, 2007, at 11:47 AM, David Lancia, Jr wrote:
> Hi Nathan -
>
> Hopefully this will be a detailed enough description to help you
> understand the problem.
>
> As part of the electrostatics calculation in a simulation, we
> currently use a pre-calculated PBE grid around a protein. This grid
> is currently calculated using the PBE solver in Congen with Amber
> parameters for radius, charge and hydrogen locations. The simulation
> expects the grid to have the same spacing in the X, Y and Z
> dimensions. It does not require the same number of spacings in each
> dimension.
>
> While Congen is technically free, it is no longer developed or
> supported and we feel it is time to attempt to replace it. The
> obvious replacement to us was APBS because of the active development
> and support community.
>
> We have developed a converter that allows us to use the APBS grid as a
> drop in replacement for the Congen grid in the simulation. However,
> due to Congen being considered the gold standard for us for so long,
> we felt it was necessary to try to quantify the differences in the two
> grids before trusting the final simulation results.
>
> The method that is failing for us at the moment is a direct comparison
> of the grid point values in the two grids. We have been able to set
> the origins the same and have the same grid size. We are using AMBER
> parameters in both Congen and APBS (via PDB2PQR). On average, the
> points near the edge of the grid differ by 10-15% and the points
> inside of the protein and near the surface can differ by 300-500%
> (worst case is a value difference of 680 (not percent)). We have
> determined this by randomly picking and comparing values and also by
> visualizing the entire 3D grids and then visualizing the differences
> between the two. I do not expect the values to match exactly due to
> differences in the solver and some minor differences in the AMBER
> parameters (I have a question for the PDB2PQR mailing list regarding
> this) but we thought we should be able to find a consistent difference
> (10-15% would be acceptable for us). As a note, the field outside
> the protein and at the surface is the most important to us, it appears
> that outside the protein we get results that are within our tolerance
> but the surface values are not acceptable without at least a better
> idea of why they could be so different.
>
> To answer some of your remaining questions:
>
> We have tried the comparison with 0mM ionic strength and this is how
> we are able to get the closest agreement of the two grids. I was
> concerned about the ION parameter so we were only going to test that
> once everything else came close to agreeing.
>
> I can forward APBS input files shortly. They need to be cleaned up at
> the moment to the version the provided us what we are considering our
> best results so far.
>
> Thanks again for any help.
>
> David
>
> On 4/8/07, Nathan Baker <ba...@cc...> wrote:
>> Hi David --
>>
>>> I have a program that currently depends on the PBE grid generated by
>>> Congen (http://www.congenomics.com/congen/congen_toc.html). While
>>> Congen is technically free, I would much rather use a system that is
>>> actively developed and supported. I have tried every combination of
>>> APBS parameters and have even modified the Congen parameters we have
>>> used for years to try to get the two grids to converge and this has
>>> proved almost futile. I am looking for any help or suggestions
>>> as to
>>> what to try
>>
>> Can you describe what quantities you're trying to get agree (e.g.,
>> solvation energy vs. pointwise error in potential vs. mean-squared
>> error in potential, etc.)?
>>
>>> First, has anyone possibly gone thru this or similar transition
>>> before
>>> and can you make any recommendations?
>>
>> There have been lots of comparisons of APBS with other PB solvers but
>> this is the first time I've seen a comparison with Congen.
>>
>>> 1. The ION parameter in APBS: In Congen, we set an IONSTR
>>> parameter
>>> to a concentration of 0.15 M. There is no ability to specify a
>>> charge
>>> or radius, so I do not know how to translate this into the APBS
>>> parameter because the ION parameter requires a charge, concentration
>>> and radius.
>>
>> I'd just use
>>
>> ion 1.0 0.150 2.0
>> ion -1.0 0.150 2.0
>>
>> since the 2.0 A radius is fairly standard. This will give you 150 mM
>> ionic strength conditions.
>>
>> However, have you tried the comparison with Congen with 0 mM ionic
>> strength?
>>
>>> 2. The fine grid to course grid relationship: Due to the nature of
>>> our legacy software, I need the grid to be the same spacing in each
>>> direction (0.8A but I am willing to change the size if that will
>>> help
>>> me). I have tried using fg_manual and using a single gird.
>>
>> The mg-manual should give you the best control. Note that grid
>> spacing (h) is related to the number of grid points (n) and the grid
>> length (L) by the relationship
>>
>> h/L = n - 1
>>
>>> I have
>>> tried fg_auto and forcing the size of the fine grid to be (# grid
>>> points * 0.8A) both with some success but nothing great. I have
>>> noticed that changing the course grid even slightly can impact the
>>> values in the PBE grid by orders of magnitude.
>>
>> Really? What are the sizes of your coarse and fine grids? This is
>> very strange behavior.
>>
>>> Is there an optimal
>>> ratio of fine grid to course grid?
>>
>> Yes, but mg-auto changes the number of focusing steps to take care of
>> this for you.
>>
>>> Or is there a better way to get a
>>> fixed width grid?
>>
>> Some more specifics about your problem (desired grid lengths, example
>> APBS input files, etc.) would help with this.
>>
>> Thanks,
>>
>> Nathan
>>
>>> Thanks.
>>> David
>>>
>>>
>>> Below are the parameters we use with Congen in case it helps anyone:
>>>
>>> pbe setup solvent 78.000000 interior 4.000000 all end -
>>> boundary adebye ionstr 0.15 -
>>> grid 0.800000 temp 300.0 water 1.4 stern 2.0 -
>>> margin 1 xdim 85 ydim 66 zdim 57 -
>>> epsave harm molsurf subdivision 5 spill uniform -
>>> smooth type grid points 15 weight constant offgrid solvent end -
>>> all
>>>
>>> --------------------------------------------------------------------
>>> --
>>> ---
>>> Take Surveys. Earn Cash. Influence the Future of IT
>>> Join SourceForge.net's Techsay panel and you'll get the chance to
>>> share your
>>> opinions on IT & business topics through brief surveys-and earn cash
>>> http://www.techsay.com/default.php?
>>> page=join.php&p=sourceforge&CID=DEVDEV
>>> _______________________________________________
>>> apbs-users mailing list
>>> apb...@li...
>>> https://lists.sourceforge.net/lists/listinfo/apbs-users
>>
>> --
>> Associate Professor, Dept. of Biochemistry and Molecular Biophysics
>> Center for Computational Biology, Washington University in St. Louis
>> Web: http://cholla.wustl.edu/
>>
>>
>>
>
> ----------------------------------------------------------------------
> ---
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to
> share your
> opinions on IT & business topics through brief surveys-and earn cash
> http://www.techsay.com/default.php?
> page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> apbs-users mailing list
> apb...@li...
> https://lists.sourceforge.net/lists/listinfo/apbs-users
|